Retail Gaze: A Dataset for Gaze Estimation in Retail Environments

Abstract

The concept of gaze object estimation predicts a bounding box that a person looks steadily. It is a applicable and contemporary technique in the retail industry. However, the existing datasets for gaze object prediction in retail is limited to controlled environments and do not consider retail product category area segmentation annotations. This paper proposes Retail Gaze, a dataset for gaze estimation in real-world retail environments. Retail Gaze is composed of 3,922 images of individuals looking at products in a retail environment, with 12 camera capture angles. Furthermore, we use state-of-the-art gaze estimation models to benchmark the Retail Gaze dataset and comprehensively analyze the results obtained.